Post on 05-Jul-2020
On Users: User Modeling based on Social Multimedia Activity (SMA)
Jitao SangMultimedia Computing Group
National Lab of Pattern Recognition, Institute of AutomationChinese Academy of Sciences
Part II
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Multimedia Analysis
Jaguar (animal)
Jaguar (car)
?
Semantic gap
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Multimedia Analysis User Modeling
Jaguar (animal)
Jaguar (car) luxury car
Occupation:animal
photographer
Interest:
Semantic gap
Intentgap
Generalized User Models
…
demographic model(age, gender, occupation) interest model
(politics, music, sports)
mobility model(Point-of-Interest)
social status model(friends, influence, propagation)
consuming model(electronics, beauty, clothing, )
emotion model(optimistic, positive negative)
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Shortage of User Information
Registration: not troubling to provide the details.
这家伙很懒,什么都没留下…
Choosing from lists: the taxonomy is arbitrary.
Privacy issues.
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Shortage of User Information
Registration: not troubling to provide the details.
这家伙很懒,什么都没留下…
Choosing from lists: the taxonomy is arbitrary.
Privacy issues.
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Out of the most 190,000 active users on Google+..
The ratio of users providing information
0.385
0.1236
0.2248
0
0.1
0.2
0.3
0.4
0.5
gender birthday marriagestatus
Extensive Social Multimedia Activities
Social Multimedia Activities
Tweets
check-in history
User Models
Photo collection
SNS posts
Favorite videos
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Categorization of Related Work
Social Status
Demographics
Interests
Others
[Hu et al. 2007; Jones et al. 2007; Otterbacher 2010; Pennacchiotti and Popescu 2011; Ying et al. 2012; Bi et al. 2013; Fang et al. 2014a]
[Koren 2010; Xiong et al. 2010; Koenigstein et al. 2011; Bennett et al. 2012; Yuan et al. 2013; Deng et al. 2014]
[Anagnostopoulos et al. 2008; Crandall et al. 2008; Xiang et al. 2010; Zhuang et al. 2011; Sang and Xu 2012; Fang et al. 2014b]
Mobility modelEmotion Consuming model
[Li et al. 2012; Yamaguchi 2013; Ahmed et al. 2013]
[Tang et al. 2012; Damian et al. 2013; Gao et al. 2014] [Zhang and Pennacchiotti 2013; Zhang et al. 2014]
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Social Status
Demographics
Interests
Others
Demographics Modeling from SMA
[Hu et al. 2007; Jones et al. 2007; Otterbacher 2010; Pennacchiotti and Popescu 2011; Ying et al. 2012; Bi et al. 2013; Fang et al. 2014a]
[Fang et al. 2014a] Quan Fang, Jitao Sang, and Changsheng Xu. UserCube: Exploiting Interaction with Multimedia Information for Relational User Attribute Inference. Submitted for publication.
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Background: Demographic Attribute Inference
Social Multimedia Activities
Tweets
check-in history
Demographic Attributes
Photo collection
SNS posts
Favorite videos
gender
age
ethnicity
occupation
User attributes are predicted independently.
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Motivation: Attributes are Connected
Occupation v.s. Interests
( Statistics based on 100 million Google+ users. )
Age v.s. Occupation
Marriage v.s. Occupation
User attributes have positive or negative intra-relations.
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Relational User Attribute Inference
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Separate SVM classifier training for each user feature:SVM-Face
SVM-ProfilePhoto
SVM-PostPhoto
SVM-unigram
SVM-sociolinguisticSVM-topic-based
Train
SVM-Face
SVM-ProfilePhoto
SVM-PostPhoto
SVM-unigram
SVM-sociolinguistic
SVM-topic-based
Fusion Stacked SVM
Relational User Attribute Inference
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Stacked SVM classifier fusion for individual attribute estimation :
Xℋfemale age<30 unmarried
0.83 0.65 0.91
Relational Latent SVM framework for enhancement
User feature vector
Auxiliary attributes Target
attribute
Stacked SVM model for each attribute
Attribute relations
Relational User Attribute Inference
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Define 6 types of attributes and their optional values:
Experiments: Attribute Example
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Experiments: Attribute Inference Evaluation
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
The derived user attribute relations:
Experiments: Attribute Relation Results
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Experiments: Attribute Relation Results
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
(a) (b)
the derived user attribute relationsthe attribute relation In the labeled dataset
Gender v.s. else
Occupation v.s. else
Experiments: Attribute Relation Results
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Structured query Ranked results
Application: Structural Attribute-based User Retrieval
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Extensions
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Attribute-based user retrieval: Formulated as a ranking problem; Consider social context (graph) information.
The observed attribute relation as supervision: First refine the observed attribute relation matrix; Fix the attribute relation as supervision, to improve attributeinference performance.
User Interest Modeling from SMA
Social Status
Demographics
Interests
Others
[Koren 2010; Xiong et al. 2010; Koenigstein et al. 2011; Wang et al. 2012; Bennett et al. 2012; Yuan et al. 2013; Deng et al. 2014]
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
User Interest Modeling: Dynamics & Context
[Wang et al. 2012] Xinxi Wang, David Rosenblum, Ye Wang: Context-aware mobile music recommendation for daily activities. ACM Multimedia 2012: 99-108. ( National University of Singapore )
Girlfriend Sleepsong
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
[Wang et al. 2012] Xinxi Wang, David Rosenblum, Ye Wang: Context-aware mobile music recommendation for daily activities. ACM Multimedia 2012: 99-108.
Running
Sleeping
… … … …
Audio content analysis Sensor based activity detection Personalization and adaptation
User Interest Modeling: Dynamics & Context
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
[Wang et al. 2012] Xinxi Wang, David Rosenblum, Ye Wang: Context-aware mobile music recommendation for daily activities. ACM Multimedia 2012: 99-108.
Activities
Ranked songs list
PlaybackControls
User Interest Modeling: Dynamics & Context
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
User Interest Modeling: Life Styles
[Yuan et al. 2013] Nicholas Jing Yuan, Fuzheng Zhang, Defu Lian, Kai Zheng, Siyu Yu, Xing Xie, We Know How You Live: Exploring the Spectrum of Urban Lifestyles. COSN 2013. ( Microsoft Research Asia )
Jiepang
Dianping
check-in
tweet
movie
music book
events restaurant
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
life-style:1-3-7
User Interest Modeling: Life Styles
[Yuan et al. 2013] Nicholas Jing Yuan, Fuzheng Zhang, Defu Lian, Kai Zheng, Siyu Yu, Xing Xie, We Know How You Live: Exploring the Spectrum of Urban Lifestyles. COSN 2013. ( Microsoft Research Asia )
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
[Yuan et al. 2013] Nicholas Jing Yuan, Fuzheng Zhang, Defu Lian, Kai Zheng, Siyu Yu, Xing Xie, We Know How You Live: Exploring the Spectrum of Urban Lifestyles. COSN 2013.
User Interest Modeling: Life Styles
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
[Yuan et al. 2013] Nicholas Jing Yuan, Fuzheng Zhang, Defu Lian, Kai Zheng, Siyu Yu, Xing Xie, We Know How You Live: Exploring the Spectrum of Urban Lifestyles. COSN 2013.
User Interest Modeling: Life Styles
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
[Yuan et al. 2013] Nicholas Jing Yuan, Fuzheng Zhang, Defu Lian, Kai Zheng, Siyu Yu, Xing Xie, We Know How You Live: Exploring the Spectrum of Urban Lifestyles. COSN 2013.
User Interest Modeling: Life Styles
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Social Status Modeling from SMA
Social Status
Demographics
Interests
Others
[Anagnostopoulos et al. 2008; Crandall et al. 2008; Xiang et al. 2010; Zhuang et al. 2011; Sang and Xu 2012; Fang et al. 2014b]
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
[Sang and Xu 2012] Jitao Sang, and Changsheng Xu. Right Buddy Makes the Difference: an Early Exploration of Social Relation Analysis in Multimedia Applications. ACM Multimedia 2012.
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Psychology
Social Science
Human Dynamics for persuasion and stress
Influence is Quantitative
Mechanism underlying Homophily:
Influence is Qualitative
Information flow and social network evolution
6
Social Multimedia ComputingAffection on behaviors, preferences or decisions
Is influence Quantitative orQualitative?
Background: Understanding Social Influence
Motivation: Social Influence is Topic-sensitive
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
techfashion
travel
fashiontech
travel
tech
fashion
travel
Milan fashion showTom
Emily
Jason
Target user Expertise
Bob
Tahiti
Data Analysis: User Interest Evolvement
Topic distribution similarity for one pair of user
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Data Analysis: User Interest Evolvement
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Data Analysis: User Interest Evolvement
1
User A’s action and topic interest is influenced by user B (contact user).
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Assumption: UGC Generative Process
User A’s action and topic interest is influenced by user B (contact user).
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
User interest evolvement data analysis:
Two ways to uploading and tagging:
Innovative: created based on own interest Influenced: affected by contact users
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Solution: Multi-modal Topic-sensitive Influence Model (mmTIM)
• Observations– Contact network – User annotated tags – User uploaded images
travel, fashion, portrait, travel,…
…
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
= Bob={ , , }Tom Emily Jason
1) sampling switch variable
2) sampling topic
3) sampling word
0 1 (⋅)0 1 1) sampling
switch variable= =
2) sampling topic
(⋅) =(⋅)topic1 topic3
Topic 2
Topic 3
Topic 1
Topic 4
=Tom
Bob
(⋅)topic1 topic3
Topic 2
Topic 3
Topic 1
=Bob
Topic 4 = =
Distribution Candidates
ΩΩΦ Tahiti
Topic 2 Topic 3Topic 1 Topic 4
0
Topic 1
1
Topic 1
Tom Emily
Jason Tom
(⋅)Topic 1
1) sampling switch variable
2) sampling topic
3) sampling word
0 1 (⋅)0 1 1) sampling
switch variable= =
2) sampling topic
(⋅) =(⋅)topic1 topic3
Topic 2
Topic 3
Topic 1
Topic 4
=Tom
Bob
(⋅)topic1 topic3
Topic 2
Topic 3
Topic 1
=Bob
Topic 4
(⋅)Topic 1
= =
Distribution Candidates
ΩΩΦ
Gibbs Sampling
21
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Input Output
( )( )( )Initialize (⋅)
(⋅)Count Compute Sample ( )( )( )
( | , , )( | , , )( | , , )
Loop
ΨΦΩ
Gibbs Sampling
Topic-word distribution
User-topic distribution
Topic-sensitive influence strength
Φ , = , , + + | |Ω , = , , , 1, + , , , 1, +, , 1 + , , 1 +Ψ , = , , , , , 0, + , , , , , 0, +, , , 0, + , , , 0, + | |
(⋅)(⋅)
Experiments
Dataset: 3,372 users (crawl their contact relationship) 30,108 unique tags 124,099 uploaded pictures 5,000 MSER visual words
#Topic = 20
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Illustration of Discovered Topics:
Topic #2travel triplandscapevacation architecture
0.01433 0.01163 0.00867 0.00681 0.00645
0.3757 0.3453 0.2657 0.2481 0.1755
Topic #13fashion dressmodelportrait style0.01213 0.00702 0.00552 0.00486 0.00461
0.2627 0.2443 0.2015 0.1578 0.1204
Experiments: Case Study
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
24
Topic #2
Topic #13
User
95638716@N00
Topic
#2
#13
Topic distribution
Favorite image
Most Influencial Contact User
Topic distribution
#follower: 504
Uploaded image Tag cloud60045418@N00
Topic distribution
Favorite image
53611153@N00 #2
#13
Uploaded image Tag cloud42759791@N00
#follower: 176Topic distribution
95386698@N00 Uploaded image Tag cloud
Topic distribution
#follower: 101
Topic distribution
23548413@N00
#follower: 373
Uploaded image Tag cloud
95638716@N00
600455418@N00
5361153@N00
42759791@N00
Experiments: Case Study
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Application 1: Personalized Image Retrieval
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
influence
Query topic 1
topic 2
topic 3
Topic space
ranked resultsTopic-sensitive Influence
Query-Adaptive influence
Document set
…
Personalized rank list
Basic idea:Social-related users’ preference can help understand the searcher’s preference.
Query
#2Topic #1
#13
#3#4#5#6#7#8
#20
……
36
Application 2:social relation
labeling
travel fashionarticulated network
behavioral network
(explicit relation)sports
(implicit relation)
Application 3: Social Media Marketing
AB
54
?
Topic-aware social multimedia marketing:
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Other User Modeling from SMA
Social Status
Demographics
Interests
OthersMobility modelEmotion Consuming model
[Li et al. 2012; Zheng et al. 2013; Ahmed et al. 2013]
[Tang et al. 2012; Damian et al. 2013; Gao et al. 2014] [Zhang and Pennacchiotti 2013a; Zhang and Pennacchiotti 2013b; Zhang et al. 2014]
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
User Mobility Pattern Modeling
[Zheng et al. 2012] Yan-Tao Zheng, Zheng-Jun Zha, Tat-Seng Chua: Mining Travel Patterns from GeotaggedPhotos. ACM TIST 2012. ( National University of Singapore )
a tourist movement trajectory
Tourist travel trails in Paris
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
[Zheng et al. 2012] Yan-Tao Zheng, Zheng-Jun Zha, Tat-Seng Chua: Mining Travel Patterns from GeotaggedPhotos. ACM TIST 2012.
Significant traffic transition pattern among Region of Attractions, in Paris
User Mobility Pattern Modeling
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
[Gao et al. 2014] Huiji Gao, Jalal Mahmud, Jilin Chen, Jeffrey Nichols, Michelle X. Zhou: Modeling User Attitude toward Controversial Topics in Online Social Media. ICWSM 2014.
User Emotion Modeling
Sentiment, opinion, and action are inter-related:
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015( Arizona State University & IBM Research)
User Emotion Modeling
between item and topic between opinion and topic
between sentiment and opinion
[Gao et al. 2014] Huiji Gao, Jalal Mahmud, Jilin Chen, Jeffrey Nichols, Michelle X. Zhou: Modeling User Attitude toward Controversial Topics in Online Social Media. ICWSM 2014.
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
[Zhang and Pennacchiotti 2013b] Yongzheng Zhang, Marco Pennacchiotti: Recommending branded products from social media. RecSys 2013
User Consuming Pattern Modeling
[Zhang and Pennacchiotti 2013a] Yongzheng Zhang, Marco Pennacchiotti: Predicting purchase behaviors from social media. WWW 2013. ( Ebay )
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
[Zhang and Pennacchiotti 2013b] Yongzheng Zhang, Marco Pennacchiotti: Recommending branded products from social media. RecSys 2013
User Consuming Pattern Modeling
[Zhang and Pennacchiotti 2013a] Yongzheng Zhang, Marco Pennacchiotti: Predicting purchase behaviors from social media. WWW 2013.
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
Summary: User Modeling from SMA
Social Status
Demographics
Interests
Mobility Emotion Consuming Model
MMM 2015 Tutorial (Part II: On Users) – Jitao Sang Jan.5, 2015
References (1)[Ahmed et al. 2013] Amr Ahmed, Liangjie Hong, and Alexander J. Smola. Hierarchical geographical modeling of user locations from social media posts. WWW 2013.[Anagnostopoulos et al. 2008] Aris Anagnostopoulos, Ravi Kumar, and Mohammad Mahdian. Influence and correlation in social networks. KDD 2008.[Bennett et al. 2012] Paul N. Bennett, Ryen W. White, Wei Chu, Susan T. Dumais, Peter Bailey, Fedor Borisyuk, and Xiaoyuan Cui. Modeling the impact of short- and long-term behavior on search personalization. SIGIR 2012.[Bi et al. 2013] Bin Bi, Milad Shokouhi, Michal Kosinski, and Thore Graepel. Inferring the demographics of search users: social data meets search queries. WWW 2013.[Crandall et al. 2008] David J. Crandall, Dan Cosley, Daniel P. Huttenlocher, Jon M. Kleinberg, and Siddharth Suri. Feedback effects between similarity and social influence in online communities. KDD 2008.[Damian et al. 2013] Borth, Damian, Rongrong Ji, Tao Chen, Thomas Breuel, and Shih-Fu Chang. "Large-scale visual sentiment ontology and detectors using adjective noun pairs.“ ACM Multimedia 2013.[Deng et al. 2014] Zhengyu Deng, Ming Yan, Jitao Sang, and Changsheng Xu. Twitter is Faster: Personalized Time-aware Video Recommendation from Twitter to YouTube. TOMM 2014.[Fang et al. 2014a] Quan Fang, Jitao Sang, and Changsheng Xu. UserCube: Exploiting Interaction with Multimedia Information for Relational User Attribute Inference. Submitted for publication.
References (2)[Fang et al. 2014b] Quan Fang, Jitao Sang, Changsheng Xu, and Yong Rui. Topic-Sensitive Influencer Mining in Interest-Based Social Media Networks via Hypergraph Learning. TMM 2014.[Gao et al. 2014] Huiji Gao, Jalal Mahmud, Jilin Chen, Jeffrey Nichols, and Michelle Zhou. "Modeling User Attitude toward Controversial Topics in Online Social Media." ICWSM 2014.[Hu et al. 2007] Jian Hu, Hua-Jun Zeng, Hua Li, Cheng Niu, and Zheng Chen. Demographic prediction based on user’s browsing behavior. WWW 2007.[Jones et al. 2007] Rosie Jones, Ravi Kumar, Bo Pang, and Andrew Tomkins.”I know what you did last summer”: query logs and user privacy. CIKM 2007.[Koenigstein et al. 2011] Noam Koenigstein, Gideon Dror, and Yehuda Koren. Yahoo! music recommendations: modeling music ratings with temporal dynamics and item taxonomy. RecSys 2011.[Koren 2010] Yehuda Koren. Collaborative filtering with temporal dynamics. Commun. ACM 2010.[Otterbacher 2010] Jahna Otterbacher. Inferring gender of movie reviewers: exploiting writing style, content and metadata. CIKM 2010.[Pennacchiotti and Popescu 2011] Marco Pennacchiotti and Ana-Maria Popescu. Democrats, republicans and starbucks afficionados: user classification in twitter. KDD 2011.[Sang and Xu 2012] Jitao Sang, and Changsheng Xu. Right Buddy Makes the Difference: an Early Exploration of Social Relation Analysis in Multimedia Applications . ACM Multimedia 2012.
References (3)[Tang et al. 2012] Jie Tang, Yuan Zhang, Jimeng Sun, Jinhai Rao, Wenjing Yu, Yiran Chen, and Alvis Cheuk M. Fong. "Quantitative study of individual emotional states in social networks.“ TAC 2012.[Wang et al. 2012] Xinxi Wang, David Rosenblum, Ye Wang: Context-aware mobile music recommendation for daily activities. ACM Multimedia 2012: 99-108.[Xiang et al. 2010] Rongjing Xiang, Jennifer Neville, and Monica Rogati. Modeling relationship strength in online social networks. WWW 2010.[Xiong et al. 2010] Liang Xiong, Xi Chen, Tzu-Kuo Huang, Jeff G. Schneider, and Jaime G. Carbonell. Temporal Collaborative Filtering with Bayesian Probabilistic Tensor Factorization. SDM 2010.[Yamaguchi et al. 2013] Yuto Yamaguchi, Toshiyuki Amagasa, Hiroyuki Kitagawa: Landmark-based user location inference in social media. COSN 2013[Ying et al. 2012] Josh Jia-Ching Ying, Yao-Jen Chang, Chi-Min Huang, and Vincent S Tseng. Demographic prediction based on users mobile behaviors. Mobile Data Challenge 2012.[Yuan et al. 2013] Nicholas Jing Yuan, Fuzheng Zhang, Defu Lian, Kai Zheng, Siyu Yu, Xing Xie, We Know How You Live: Exploring the Spectrum of Urban Lifestyles. COSN 2013.[Zhang and Pennacchiotti 2013] Yongzheng Zhang, Marco Pennacchiotti: Recommending branded products from social media. RecSys 2013.
References (4)[Zhang et al. 2014] Fuzheng Zhang, Nicholas Jing Yuan, Defu Lian, and Xing Xie, Mining Novelty Seeking Trait Across Heterogeneous Domains, WWW 2014.[Zheng et al. 2012] Yan-Tao Zheng, Zheng-Jun Zha, Tat-Seng Chua: Mining Travel Patterns from Geotagged Photos. ACM TIST 2012.[Zhuang et al. 2011] Jinfeng Zhuang, Tao Mei, Steven C. H. Hoi, Xian-Sheng Hua, and Shipeng Li. Modeling social strength in social media community via kernel-based learning. ACM Multimedia 2011.
Outline
Introduction (15’)
Part I – From Users: User-perceptive Multimedia Analysis (1h)
Break (15’)
Part II - On Users: User Modeling based on Social Multimedia Activity (35’)
Part III: User-centric Cross-network Social Multimedia Computing (35’)
Conclusion & QA (20’)
MMM 2015 Tutorial (Introduction) – Jitao Sang Jan.5, 2015